Make the formatter smarter about mutli-line steps

src/expr-parser.ts

@@ -10,7 +10,7 @@ type SimpleLocation = { start: { offset: number }; end: { offset: number } };
 type BareText = {
  name: 'text';
  contents: string;
- location: { start: { offset: number }; end: { offset: number } };
+ location: SimpleLocation;
 }; // like TextNode, but with less complete location information
 type ProsePart = FragmentNode | BareText;
 type List = {

src/formatter/ecmarkdown.ts

@@ -34,6 +34,32 @@ async function printListNode(
  return output;
 }
 
+function commonIndent(lines: string[]) {
+ if (lines.length === 0) {
+ return '';
+ }
+ let common = lines[0].match(/^[ \t]+/)?.[0];
+ if (common == null) {
+ return '';
+ }
+ for (let i = 1; i < lines.length; ++i) {
+ const line = lines[i];
+ let j = 0;
+ for (; j < line.length && j < common.length; ++j) {
+ if (common[j] !== line[j]) {
+ break;
+ }
+ }
+ if (j <= 0) {
+ return '';
+ }
+ if (j < common.length) {
+ common = common.slice(0, j);
+ }
+ }
+ return common;
+}
+
 async function printStep(
  source: string,
  item: OrderedListItemNode | UnorderedListItemNode,
@@ -48,7 +74,14 @@ async function printStep(
  .join(', ');
  output.appendText(`[${joined}] `);
  }
- const contents = await printFragments(source, item.contents, indent + 1);
+
+ const stepSource = source.substring(
+ item.location.start.offset,
+ item.contents[item.contents.length - 1].location.end.offset,
+ );
+ const stepIndent = commonIndent(stepSource.split('\n').slice(1));
+
+ const contents = await printFragments(source, item.contents, indent + 1, stepIndent);
  // this is a bit gross, but whatever
  contents.lines[0] = contents.lines[0].trimStart();
  output.append(contents);
@@ -69,6 +102,8 @@ export async function printFragments(
  source: string,
  contents: FragmentNode[],
  indent: number,
+ // for steps which are split over multiple lines, this is the common indent among them
+ commonIndent: string = '',
 ): Promise<LineBuilder> {
  const output = new LineBuilder(indent);
  let skipNextElement = false;
@@ -82,7 +117,7 @@ export async function printFragments(
  // so don't even bother
  const { start, end } = node.location;
  const originalText = source.substring(start.offset, end.offset);
- output.append(printText(originalText, indent));
+ output.append(printText(originalText, indent, commonIndent));
  break;
  }
  case 'comment': {

src/formatter/line-builder.ts

@@ -58,7 +58,7 @@ export class LineBuilder {
  if (text === '') {
  return;
  }
- this.last = ' '.repeat(this.indent);
+ this.last += ' '.repeat(this.indent);
  }
  this.last += allowMultiSpace ? text : text.replace(/ +/g, ' ');
  }
@@ -80,7 +80,6 @@ export class LineBuilder {
  }
 
  linebreak(): void {
- // console.log('linebreak');
  if (this.isEmpty()) {
  this.firstLineIsPartial = false;
  return;
@@ -121,6 +120,6 @@ export class LineBuilder {
  // when firstLineIsPartial, we don't indent the first line
  return false;
  }
- return this.last === '';
+ return /^ *$/.test(this.last);
  }
 }

src/formatter/text.ts

@@ -22,7 +22,7 @@ function isBadNumericReference(codePoint: number) {
  return false;
 }
 
-export function printText(text: string, indent: number): LineBuilder {
+export function printText(text: string, indent: number, commonIndent: string = ''): LineBuilder {
  const output: LineBuilder = new LineBuilder(indent);
  if (text === '') {
  return output;
@@ -64,25 +64,42 @@ export function printText(text: string, indent: number): LineBuilder {
  const leadingSpace = text[0] === ' ' || text[0] === '\t';
  const trailingSpace = text[text.length - 1] === ' ' || text[text.length - 1] === '\t';
 
- const lines = text.split('\n').map(l => l.trim());
+ const lines = text.split('\n').map((l, i) => {
+ if (i === 0 || commonIndent === '' || !l.startsWith(commonIndent)) {
+ return l.trim();
+ }
+ const withoutIndent = l.substring(commonIndent.length);
+ const moreIndent = withoutIndent.match(/^ +/)?.[0];
+ if (moreIndent) {
+ return { indent: moreIndent, line: l.trim() };
+ }
+ return l.trim();
+ });
 
  if (leadingSpace) {
  output.appendText(' ');
  }
  if (lines.length === 1) {
  if (lines[0] !== '') {
- output.appendText(lines[0]);
+ output.appendText(lines[0] as string);
  if (trailingSpace) {
  output.appendText(' ');
  }
  }
  return output;
  }
- for (let i = 0; i < lines.length - 1; ++i) {
- output.appendText(lines[i]);
- output.linebreak();
+ for (let i = 0; i < lines.length; ++i) {
+ const line = lines[i];
+ if (typeof line === 'string') {
+ output.appendText(line);
+ } else {
+ output.last = line.indent;
+ output.appendText(line.line);
+ }
+ if (i < lines.length - 1) {
+ output.linebreak();
+ }
  }
- output.appendText(lines[lines.length - 1]);
  if (trailingSpace && output.last !== '') {
  output.appendText(' ');
  }

test/formatter.js

@@ -493,6 +493,29 @@ describe('algorithm formatting', () => {
  );
  });
 
+ it('multiline steps', async () => {
+ await assertDocFormatsAs(
+ `
+ <emu-alg>
+ 1. Return the Record {
+ [[Precision]]: *"minute"*,
+ [[Unit]]: *"minute"*,
+ [[Increment]]: 1
+ }.
+ </emu-alg>
+ `,
+ dedentKeepingTrailingNewline`
+ <emu-alg>
+ 1. Return the Record {
+ [[Precision]]: *"minute"*,
+ [[Unit]]: *"minute"*,
+ [[Increment]]: 1
+ }.
+ </emu-alg>
+ `,
+ );
+ });
+
  it('nested html', async () => {
  await assertDocFormatsAs(
  `